UCCX Fail Over

[scooter817]

I have a quick question, we have redundant UCCX servers and a few moments ago the master went down and no one could log into Finesse. Shouldn't the slave have taken over and allow the user to still log into Finesse, my manager asked me why didn't this happen and I wanted to find out for him.

Eric

[Shalid Kurunnan Chalil]

Hi, 

UCCX - Finesse failover well explained by Deepak in the following post - please have a look and try out his steps. 

https://community.cisco.com/t5/contact-center/uccx-finesse-failover/td-p/3094272

update with your result and we can have look again. 

Regards,

Shalid 

Please rate useful pos

[scooter817]

Shalid

From reading the information it seems like the only way to get users to be able to log into Finesse is to have both URL published on a load balancer.

[Anthony Holloway · Accepted solution]

It depends on your version of UCCX, how the server "went down," and whether or not the Agent was already logged in or not.  It's actually pretty complicated to describe all the ways in which this can fail, but I can assure you that you do not *need* a front-end load-balancer in order to have redundancy.  For instance, you could simply publish both links to the users, and educate them to click one first, and if it fails, try the other.  Now, a front-end load-balancer is nice to have, granted, but not required.

[scooter817]

Thanks Anthony, I had no Idea that if you aren't logged into Finesse and the primary node goes down it wouldn't automatically fail you over when you try and log in.

[Anthony Holloway · Accepted solution]

Just think about if the server crashed, and you try to access it via URL in your browser. How could it redirect you? The only way you could reasonably expect it to redirect you, is if it's operational enough to be aware of it's partner server, but not operational enough to actually login an Agent. That's a very narrow window of opportunity. I'd wager, most unexpected failures wont be so kind to your users.

PS The whole idea of redirecting users to the other server is a concept which helped with the fact that only one server could login Agent at a time, and that was the Master. Since version 11.6 onward, Agents can login to either server: Master or Slave, and will no longer be redirected.

[Anthony Holloway]

Two questions:

1) What type of failure did the Master experience?

2) Where the Agents already logged into the Master at the time of the failure, or did the failure occur first, while the Agents were not logged in?